OpenKE is an open-source framework for knowledge embedding organized by THUNLP based on the TensorFlow toolkit. In OpenKE, we provide fast and stable toolkits, including the most popular knowledge representation learning (KRL) methods. The framework easy to extend and convenient to design new knowledge embedding models.
Features include:
Simple interfaces to configure various training environments and classical models.
Speed and memory optimizations for high-performance GPU training.
Efficient lightweight C++ implementations for quick deployment and multi-threading acceleration.
Pretrained embeddings of the existing large-scale knowledge graphs available for several related tasks.
Long-term maintenance to fix bugs and meet new requests.